gregh
electronics, computers and scratchbuilding
A story about two almost identical locos. One works great and I?m sure the other is JINXED.
Back in 2005 I built a 2-8-2 tank loco based on an Aristo 0-8-0 motor block (ART 29357). It proved to one of my best locos.
So in 2009 I bought another one of these motor blocks and built a 4-8-0 tender loco. This has turned out to be a DUD and I don?t know why.
Both locos use 10, NiMH cells and my own design RC using 433 MHz and Picaxe (programmable controller) based PWM controller, and sound.
The tank loco used the standard Picaxe PWM frequency of 3900 Hz and worked well. No whine or whistle from the motor and good low speed control.
The first problem: I installed the same circuit and program in the 4-8-0 tender loco and it did not work well. Wouldn?t start to move till about 6V applied. Thought it might be the FET and replaced it but no change. So I rewrote the program for a 66 Hz PWM and it worked well. No hum.
Second problem: Then I thought I?d be clever and add a low battery alarm as there was a spare Analogue to digital converter. Decided to set the alarm level to detect one bad cell = 9 cells x 1.2V = 10.8V.
Note that I don?t have this alarm on the tank loco, so maybe ignorance is bliss.
I only ran the loco for 5 minutes before the alarm went off. So onto the bench again and more measurements. Found a bad connection in the plug between tender and loco, causing a 1V drop. Fixed that, but still only ran for 10 minutes before alarm. Measured that the wiring caused another 0.5V drop at 1 amp (the measured current when working hard). Couldn?t change that, so reset the alarm level to 10.3V.
More running and only 10 minutes run before alarm, so measured the internal resistance of each battery cell. All similar at around 0.13 ohms = 1.3 ohms for the whole battery). A search on the internet says a typical figure is 0.17 ohms so that seemed OK. So I set the alarm at 10.1V (that?s 12.0V ? 0.6V in leads/plugs and 1.3V in battery at 1 amp average.) And it still gives an alarm after running for 10 minutes.
In all cases the battery volts on no load were around 12.5V which seemed fine and every cell had the same voltage on load.
Then I was writing an article on PWM for GSC and decided to measure the current using an oscilloscope to see the instantaneous current and not the average measured by a multimeter. Because of the low frequency PWM (66 Hz) the current pulse is actually about 2A peak, so the instantaneous voltage drop is 3.8V giving 8.2V.
I finally decided to change the whole battery. I had some HobbyKing cells so I tested their internal resistance at only 0.06 ohms (half of the existing ones). So I installed them.
They are certainly an improvement, but I still only get half the run time of the 2-8-2 tank. So I?m still looking for the JINX!
Back in 2005 I built a 2-8-2 tank loco based on an Aristo 0-8-0 motor block (ART 29357). It proved to one of my best locos.
So in 2009 I bought another one of these motor blocks and built a 4-8-0 tender loco. This has turned out to be a DUD and I don?t know why.
Both locos use 10, NiMH cells and my own design RC using 433 MHz and Picaxe (programmable controller) based PWM controller, and sound.
The tank loco used the standard Picaxe PWM frequency of 3900 Hz and worked well. No whine or whistle from the motor and good low speed control.
The first problem: I installed the same circuit and program in the 4-8-0 tender loco and it did not work well. Wouldn?t start to move till about 6V applied. Thought it might be the FET and replaced it but no change. So I rewrote the program for a 66 Hz PWM and it worked well. No hum.
Second problem: Then I thought I?d be clever and add a low battery alarm as there was a spare Analogue to digital converter. Decided to set the alarm level to detect one bad cell = 9 cells x 1.2V = 10.8V.
Note that I don?t have this alarm on the tank loco, so maybe ignorance is bliss.
I only ran the loco for 5 minutes before the alarm went off. So onto the bench again and more measurements. Found a bad connection in the plug between tender and loco, causing a 1V drop. Fixed that, but still only ran for 10 minutes before alarm. Measured that the wiring caused another 0.5V drop at 1 amp (the measured current when working hard). Couldn?t change that, so reset the alarm level to 10.3V.
More running and only 10 minutes run before alarm, so measured the internal resistance of each battery cell. All similar at around 0.13 ohms = 1.3 ohms for the whole battery). A search on the internet says a typical figure is 0.17 ohms so that seemed OK. So I set the alarm at 10.1V (that?s 12.0V ? 0.6V in leads/plugs and 1.3V in battery at 1 amp average.) And it still gives an alarm after running for 10 minutes.
In all cases the battery volts on no load were around 12.5V which seemed fine and every cell had the same voltage on load.
Then I was writing an article on PWM for GSC and decided to measure the current using an oscilloscope to see the instantaneous current and not the average measured by a multimeter. Because of the low frequency PWM (66 Hz) the current pulse is actually about 2A peak, so the instantaneous voltage drop is 3.8V giving 8.2V.
I finally decided to change the whole battery. I had some HobbyKing cells so I tested their internal resistance at only 0.06 ohms (half of the existing ones). So I installed them.
They are certainly an improvement, but I still only get half the run time of the 2-8-2 tank. So I?m still looking for the JINX!
